actionable knowledge

Print Publication Date: Mar 2005 Subject: Business and Management, Organizational Theory and Behaviour, Technology and Knowledge ManagementOnline Publication Date: Sep 2009 DOI: 10.1093/oxfordhb/9780199275250.003.0016

Actionable KnowledgeChris ArgyrisThe Oxford Handbook of Organization Theory

Oxford Handbooks Online

Abstract and KeywordsThe focus of this article is upon producing actionable knowledge. Propositions that are actionable are those that actors can use to implementeffectively their intentions. Actionable knowledge requires propositions that make explicit the causal processes required to produce action. Causality isthe key in implementation. One of the most powerful inhibitors of effective action is inner contradictions. Inner contradictions exist when the propositionsto act are implemented correctly. One cause of inner contradiction is the methodologies used by most normal social scientists to discover problemsand to invent solutions. These features cause the degree of seamlessness and the validity of the implementation to be reduced. The focus ondescribing reality in ways that satisfies the requirements of internal and external validity makes it less likely that attention is paid to the implementablevalidity of the propositions. This, in turn, leads to propositions that are abstract and disconnected from implementable action.Keywords: actionable knowledge, causality, inner contradictions, social scientists, implementable validity

The focus of this chapter is upon producing actionable knowledge. Propositions that are actionable are those that actors can use to implementeffectively their intentions. Effective implementation occurs when (a) a match is produced between the intentions of the actors and the actual outcomes,(b) in such a way that the outcomes persevere, and (c) without deteriorating the existing level of problem-solving effectiveness.

15.1 The Importance of Actionable KnowledgeWhy should social scientists focus upon actionable knowledge?

1. Most social scientists claim that producing knowledge that can assist in enhancing human effectiveness is an important objective for scholars.(Read, for example, in my country, the charters of the American Psychological, Sociological, and Anthropological Associations.) Many researcherswho support actionable knowledge also say that in order to produce actionable knowledge we may have to wait until the descriptive knowledgecumulates to the point that it can be used to produce actionable knowledge. In an analysis of the literature I concluded that, in the field of organizationalstudies, there is little additivity of knowledge that leads to the promises made by scholars (Argyris 1980).

Since then, I have suggested another reason why the additivity required to produce actionable knowledge is not likely to occur. The reason is that theideas in good currency about rigorous research inveigh against the production of actionable knowledge. This is especially true if the knowledge is aboutchanging the status quo and creating what some scholars call liberating alternatives. More about this below.

2. The most powerful empirical tests for theories are provided when predictions are made about changing the universe, not simply describing it. Lewinmade this point when he said that if we wished to truly understand the universe, try changing it. When we describe the universe, there is some wiggle-room for our claims. As we shall see when we strive to change the universe, we require knowledge that can be used to create something new and rare.In order to do so, we must specify causal propositions about how to create change. Moreover, as we shall see, these causal propositions must becrafted in the form that the human mind can store, retrieve, and implement in an on-line manner.

3. Most theories and much empirical research often contain gaps, inconsistencies, and inner contradictions. Some of these arise because ofincompleteness of our knowledge. The expectation is that with further empirical research, they will be discovered. The expectation is often not fulfilled.

But, what if these gaps, inconsistencies, and inner contradictions occur because we are following correctly the rules of normal science research? Whatif these rules and the norms that they create in the scholarly community help to keep us unaware of the gaps and inconsistencies and unaware that weare unaware?

I suggest that the most powerful way to surface these issues is to attempt to intervene; to state predictions about creating phenomena that do not existpresently.

4. Actionable knowledge is the knowledge that is most likely to be of help to human beings because it prescribes how they should act. The basis for thesense of competence, self-esteem, and self-efficacy is effective action (White 1959; Bandura 1986). Action is therefore at the heart of what it means tobe human.

Actionable knowledge also empowers the users. It provides them with powerful reasons to accept or reject the claims of applicability made byresearchers. The users are empowered because they can see that they have been helped or because they have good evidence to reject the claim of

Actionable Knowledge

PRINTED FROM OXFORD HANDBOOKS ONLINE (www.oxfordhandbooks.com). (c) Oxford University Press, 2013. All Rights Reserved. Under theterms of the l icence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (fordetails see Privacy Policy).Subscriber: OUP - Oxford Online %28Sales %26 Publicity%29; date: 13 April 2014

the validity of the knowledge.

15.2 Norms and Rules About Producing Scientific KnowledgeThere are four fundamental norms and rules about producing scientific knowledge that, when implemented correctly, will limit and inhibit the productionof valid actionable knowledge. They are (1) describe the universe as is, (2) be pluralistic in outlook, (3) focus on internal and external validity, all ofwhich combine to create, (4) unintentional gaps and inconsistencies.

15.2.1 The Importance of Describing the Universe as Completely and Validly as PossibleOne of the most fundamental norms of the social science community is that the task of scholars is to describe their chosen universe as completely andvalidly as possible. The test for completeness and validity is the derivation of falsifiable hypotheses that are tested empirically.

One rule that flows from this norm is that scholars should remain descriptive and avoid being normative and especially prescriptive. The latteremphases are delegated to applied researchers who are not conducting basic research.

For example, Cronbach and Suppes (1969) separate conclusion-oriented research (basic) from decision-oriented research (applied). Coleman(1972) continued and expanded the distinction by calling the former discipline research and the latter policy research. Many scholars accept thisdistinction. An excellent compilation of their positions can be found in Stringer (1982). Basic research is designed to test theories and to producegeneralizations, is open to scrutiny by the scholarly community, and does not begin with an intention to be of help or to produce implementableknowledge (Coleman 1972).

Action or applied research, on the other hand, seeks to solve client problems and to produce workable solutions that are usually not abstract (in orderto be relevant to the concrete problems of practice), are rarely subject to scholarly scrutiny, and are intended for implementation. Some would go asfar as saying that it might not be necessary or useful for such researchers to be concerned about truth. Indeed, they might find that being helpfulrequires ignoring truth (Ellis 1982).

McGrath and Brinberg (1984) have developed a model that attempts to show why many of these differences between basic and applied researchmay be symptomatic of researchers defenses. Using the validity network schema, they are able to show that basic and applied research may havedifferent pathways but that both are centrally concerned with the validity and generalizability of propositions. They suggest that both camps show aninterest in each other's biases but that they tend to deal with that interest by mentioning it last. I believe that there is much wisdom in McGrath andBrinberg's perspective. However, I hesitate to agree with their proposition that basic researchers are primarily concerned with the conceptual domainand applied researchers with the substantive domain.

Peters and Robinson (1984) conclude that both versions of applied research espouse involvement-in-change and organic research processes heavilyinfluenced by collaboration between researchers and subjects. They also conclude that the researchers who advocate the strong version tend to placea heavier emphasis upon constructivist / interactionist epistemology. The latter stresses that our understanding of the world is partially constitutive of it,and that social actors create their own histories and are capable of reflection-in-action in order to change the world as it is. The strong version, theyconclude, is emancipatory.

I believe further that research to produce actionable knowledge can and should be exposed to self-correcting procedures. It may be, as Coleman(1972) suggests, that much of the self-correcting activity in present policy research is best conducted by independent studies. I am recommending thatself-correcting procedures should be designed into studies because these procedures protect the human subjects, as well as foster the production ofvalid basic knowledge.

15.2.2 Unintended Cover-up of Gaps and InconsistenciesThe second problem with separating descriptive from normative and prescriptive research is that it inhibits the most robust testing that is possible of atheory; it unintentionally covers up the gaps and inconsistencies of a theory, and finally it places social science in the role of being the servant of thestatus quo. I should like to illustrate these claims by inquiring into the Behavioral Theory of the Firm (BTF). Cyert and March (1963) define four relationalconcepts at the core of their perspective. They are: (1) quasi-resolution of conflict, (2) uncertainty avoidance, (3) problemistic search, and (4) limitedorganizational learning. Twenty-five years later, in an introduction to a collection of his essays, March (1988) identified four key concepts of BTF, whichhe now described as heresies in the eyes of many economists and students of traditional decision-making. They are: (1) the importance of allocatingattention, (2) the omnipresence of conflict, politics, and coalition groups, (3) the conception that action involves rules that adapt to experience, ratherthan anticipatory choices, and (4) the importance of ambiguity, preferences, technology, and history.

March takes on concepts such as routines, control, loyalty, and trust. March (1981: 221) states that it is the standard operating procedures androutines that are causally responsible for much of the behavior that is observed in organizations.

Mundane rules are at the heart of organizational activity. Rules, in turn, are based upon previous experience and history (p. 224). If change is to occur,it must be clearly linked to mundane rules (p. 222).

What is the meaning of linked? One reading is that all changes, if they are to persevere, have to be linked to routines and procedures. I agree.

There could be another reading of the term linked. This reading could lead researchers to focus so heavily upon routines that they pay little attention toresearch that is aimed at non-trivial changes in the routines themselves. The focus on routines leads to reinforcing the status quo. For example, if itworks, don't fix it or question it. The routines are taken for granted. One of the most difficult learning problems organizations may face is to learn that



they are not able to learn, and that the cause of this inability is the focus on what is taken for granted namely, routines. If one claims that limitedlearning, quasi-resolution of conflict, and competitive coalition groups are omnipresent and routine, it is not difficult to conclude that they are to be takenfor granted and therefore not likely to be altered.

I turn next to the concept of control. March states that control is a fundamental process in most organizations. I agree. Then, March links control andconflict in the following way. He suggests that control systems drive conflict systems because once measures are developed to evaluate performanceand compliance they invite manipulation. Once the rules of evaluation are set, conflict of interest between the rule setters and their followers assuresthat there will be some incentives for the latter to maximize the difference between the rule setters and the rule followers. It assures that there will besome incentives for the latter to maximize the difference between their score and their effort. Any system of accounts is a road map to cheating onthem (March 1981: 220).

The reasoning appears to be as follows. Control processes are fundamental to organization; controls require the evaluation of performance. Peoplecope with such evaluations by manipulation and cheating. They create coalition groups to bring about and to protect these self-protective processes. Ifthis self-interest is not consistent with the organization's, then more manipulation and cheating will occur. This, in turn, results in more control. Theprocesses can become self-reinforcing.

March uses similar logic in discussing trust and loyalty. Trust and loyalty, he asserts are hard to find. The problem of trust is exacerbated byorganizational politics. Organizational politics is a central feature of most organizations. [T]he first principle of politics is that everyone is rational and noone can be trusted. There may be a few who can be trusted, but they are to be characterized as innocent and naive (March 1981: 219).

The consequence is that the organizations are composed of winners and losers. The players will try to look trustworthy even though they are not, inorder to be trusted by those people who might become winners. It is therefore, a palpable feature of organizational life, that organizations may bevalidly characterized as individuals and groups pursuing their own interests by the manipulation of information (March 1988: 6). It is not surprising thatMarch concludes changes in the domains of conflict and trust through the confrontation and resolution of conflict are not likely to be effective (March1988: 8). In effect, March paints a valid (I believe) picture of dysfunctionalities around issues of control, loyalty, trust, cheating. He then argues thatthese are not likely to be alterable. Not surprisingly, he and others have not conducted empirical research to test the claim.

There is a theme running throughout this reaction that also occurs throughout the chapter. The theme is illustrated by Lewin's advice that if we wish tounderstand the universe fully we should strive to change it. The discussion above provides several reasons why this advice is sound.

First, as we have seen, describing the universe, tells us little about the actions that the members of that universe used to create the universe in the firstplace. Thus, the relational concepts represent, in my view, valid descriptions of organizations. However, stopping at this description tells us very littleabout how these variables were produced in the first place.

Second, if the universe is described as being composed of relatively stable patterns, it is necessary for us to explain the causal mechanism by whichthe patterns maintain their stability. These causal mechanisms are not likely to be discovered until there is an intervention intended to change themfundamentally. Under these conditions, the causal mechanisms that create the stability will be activated.

It follows, I suggest, that making description the ultimate goal risks placing organization theory in the role of being the servant of the status quo. The riskrelated to the world of practice is that such knowledge will not help practitioners who seek to change the status quo. The risk to science is that the goalof seeking as complete descriptions of the universe is limited by design because little attention is paid to understanding how to make fundamentalchanges in the patterns.

15.2.3 The Importance of PluralismA basic norm of normal science is that researchers should focus on describing the universe of their choice. The researchers should craft their findingsin the form that they are generalizable and empirically testable. One result of this is the development of a variety of frameworks and methodologies.The norm is that scholars are free to pursue their interests in describing their chosen universe. For example, Bolman and Deal 1991; Burrell andMorgan 1979; Gill and Johnson 1991; McGrath, Martin, and Kukla 1982; Morgan 1983; Scott 1981; Van Maanen 1982; advocate a pluralisticapproach as an effective strategy for seeking truth.

There are, at least, two difficulties with this approach. The first is that since valid description and testable propositions are the two main qualifiers forvalid scholarship, most scholars focus primarily on developing their own perspectives and on being critical of competing perspectives. The result isintergroup rivalries where the critiques provide little insight into how to close gaps and inconsistencies, and even less as to how to produce actionableknowledge. For example, the humanistic-interpretive scholars assert, I believe correctly, that positivists strive to be objective. In doing so, they tend toignore the meanings held by human beings because a consideration of those meanings would require an intuitive, subjective, and empathic grasp ofthe subjects consciousness (Giddens 1976). They also tend to ignore the processes by which their subjects construct their realities (Rosen 1991).

This results in positivistic researchers distancing themselves from their subjects. Humanistic-interpretive researchers get close to their subjects,according to Van Maanen (1982), because they live among their subjects. I agree with this conclusion up to a point. The point can be defined byasking, how close is close?

Van Maanen certainly had close relationships with his subjects. Witness the rich dialogue that he reports, including candid comments by somepolicemen about their behavior toward the citizens they arrest. I think it is fair to say that Van Maanen documents that police expressed a nontrivialamount of hostility and prejudice toward the citizens they serve. He also documents how this hostility can become internalized by individuals(appearing, for example, as constant swearing) and can be the basis for a police culture that protects such individual defenses.

Van Maanen does not, as far as I can tell, conduct research to explore the causes of these individual and cultural defenses in more depth. This could



be done, for example, by developing an intervention program, intended to surface and reduce the defenses. I believe Van Maanen would not disagreewith this possibility but would maintain that intervention to change taken-for-granted behavior is not part of his research practice. The point that I ammaking is not that he must change. The point is that he and other ethnographers ought to specify the distancing that they create, just as they specifythe distancing the positivists create.

Another position often taken by field researchers who focus on naturalistic observations is that such research goes deeper than does research that isdesigned to be distant in order to be objective. Again, this position seems plausible. However, there are gaps and inconsistencies to be found, whichare illustrated by Dyer and Wilkins's (1991) critique of Eisenhardt's (1989) position on how to conduct research.

Dyer and Wilkins favor using stories to generate theory. They believe that single case stories can deliver deep insights, while those developed fromcomparative case studies are likely to be thin. The difficulty is that they craft their position so that it is not disconfirmable; indeed, it is not producible.For example, what are the properties of an effective story? What are the properties of thin insights? Dyers and Wilkins state: Although it is difficult todetermine how deep a researcher must go to generate good theory, the classic case study researchers certainly went deeper into the dynamics of asingle case than Eisenhardt advocates (p. 616). But the authors do not define what an appropriate depth is. Until this is done, how can we judgeEisenhardt's position to be wanting?

Dyer and Wilkins assert that in multi-case studies the focus is on the construct to the detriment of the context. I am in favor of focusing on the context.But those of us who support contextual research are, I suggest, responsible for defining context in a way that is not self-referential, which makes testingfeatures of it difficult. Dyer and Wilkins approvingly cite several classic case studies that tell a story (for example, Whyte 1991). Yet, as Eisenhardtshows in her reply, those studies had features that were consistent with her view of context as much if not more than with Dyer and Wilkins's view.

Finally, Rosen (1991) also suggests that ethnographers do not allow any theoretical preoccupations to decide whether some facts are more importantthan others. I agree that competent ethnographers focus on the relatively directly observable data. It is difficult to agree, however, that ethnographersdo not allow any theoretical preconceptions to decide whether some facts are more important than others. As I see it, by eschewing interventions tochange the constructed world they describe, they ignore, or do not permit to surface, the data that would arise if anyone tried to change that world.They may not let their preconceptions decide which facts are and are not important. But they allow their preconceptions to create conditions underwhich crucial facts will never arise.

Kunda is quoted by Rosen (p. 21) as saying that ethnography is the only human activity in the social sciences [that] is not divorced from the modesof experience that I consider human. I do not question that Kunda considers ethnography to be connected to what it means to be human, but thevalidity of the assertion is, in my judgment, doubtful. Kunda's study, which in my opinion is an excellent example of ethnography, misses the core ofwhat it means to be human, according to his standards. Being human means to act, to construct a world, and to engage the world in ways that engagethe actors.

Kunda observed human beings being human, but he never became human in the sense of taking action to help his subjects in the struggle of beinghuman. It was possible for Kunda to become as human as his subjects by developing interventions. Interventions are human experiments that havethe intention of constructing different virtual worlds. Engstrom (2000) suggests that ethnographers consider seriously to become more interventionist inapproach if they are to fulfill the claims that they espouse.

Another example of counterproductive rivalries is illustrated by a review of invited papers to a conference on applied communications. The paperswere selected by applied communications experts as representing some of the best and current empirical research representing the naturalistic,humanistic, interpretive (NHI) approaches. An analysis of these papers indicated that the writers were critical of the objectivist positive approach. Yet,the papers represented some puzzles. For example, the proponents of NHI speak of wholeness, choice, meaning, flexibility, open texts, and dialogue.Yet much of the research that I reviewed was in the service of compliance, closed dialogue, and closed texts. Many of the scholars questioned thenormal science concepts of causality. Yet, my analysis suggests that the reports rely upon a tacit concept of causality. The NHI scholars were criticalof the distancing produced by positivism yet, they produced their own distancing and seemed unaware that they did so (Argyris 1995).

Finally a problem arises ironically by attempts made by scholars to reduce the counterproductive consequences of academics by integrating variousapproaches. The result of these approaches is to integrate through the use of approaches that amount to a live-and-let-live approach where thestrengths of each approach is emphasized and the weaknesses are de-emphasized.

For example, Morgan and Smircich (1980) suggest a way to integrate the various paradigms. They present an analytical model that describes eachparadigm's assumptions about ontology, human nature, epistemology, metaphors, and research methods. The resulting thirty-box table is organizedaround the concepts of inter-pretivist and functionalist paradigms, which, in turn, are conceptualized along a continuum ranging from subjectivist (forthe former) to objectivist (for the latter).

A first reaction to the table is that it makes good sense. There are indeed paradigms that are primarily subjectivist and others that are primarilyobjectivist. However, a second reaction is that the table does not accurately characterize the assumptions regarding ontology, human nature,epistemology, metaphors, and research methods if intervention and implementation are to be taken seriously. Those who conduct such research findthat almost all the categories are relevant. For example, in an organizational study, it is shown that the pattern of the director group's defensive routinesrepresents reality as a social construct. It is implemented in a realm of symbolic discourse. It is a contextual field of information. It is a representation ofconcrete processes and structures in the sense that these processes and structures exist out there. It coerces different directors to behave in similarmanners (Argyris 1993).

The assumptions about human nature illustrated in this research also range from the extremes of subjectivism to those of objectivism. For example,the participants were symbol creators, symbol users, and actors using symbols. The intervention research depended heavily on understanding themas information processors with limited capacity to deal with environmental complexity. Many examples were described where the same individuals



were primarily adapters and responders before and after the interventions.

Finally, the research methods ranged from explorations of pure subjectivity, to script and symbolic analysis, to contextual analysis, to historicalanalysis of the director group, and to the design and execution of many experiments. Moreover, the entire research project occurred over a period offive years before publication and it continues. History became increasingly important as the interventions became cumulative, but so did the testingand experimenting used to assess the extent to which the new pattern contained processes and structures that coerced action in the service oflearning and of reducing defensive routines at all levels.

When individuals, groups, intergroups, or organizations are studied under conditions where interventions are an integral part of the activity, I suggestthat both subjectivist and objectivist assumptions will always be relevant. For example, the interventionists had to help the directors become aware ofhow they constructed reality (subjectivist). They then helped them see how these constructions led to a pattern that was out there because they hadplaced it out there, through actions designed to inhibit. This, in turn, required that they map contexts and study systems and processes. Finally, this ledthe interventionists to help the directors see how they had created a world where a positivist stance was both necessary and counterproductive.

When the subjects were helped to learn a new pattern, it was done by using metaphors and to keep side-by-side the metaphors of theater and cultureand the metaphors of cybernetics and organisms. The directors were also helped to design and implement many experiments in order to create overtime a new pattern that was out there and that coerced the new behavior.

When the objective is to produce actionable knowledge, the strategy is not consistent with the pluralistic strategy of using several different perspectivesrelatively independently of each other. For example, Hassard (1991) has described a study in which four different paradigms (functionalist,interpretive, radical structuralist, and radical humanist) were used to study features of an organization. Each study was conducted consistently witheach perspective's assumptions regarding ontology, epistemology, human nature, and methodology. Each produced different descriptions of differentfeatures of the organization.

Moreover, there was no attempt reported to produce actionable knowledge relevant to changing the status quo. The exercise met the needs of theresearchers, who wanted to see what the different perspectives would produce, and the subjects interests were subordinated to the researchersinterests. The same is true of the analysis made by Allison (1971) of the Cuban missile crisis.

The researchers also produced, by design, defensive routines in their relationships with the subjects. They withheld their interest in conducting theresearch with four different paradigms, because they feared that they might not gain access if they revealed their intentions fully. In my view, theybypassed embarrassment and threat and covered up the bypass.

Donaldson (1985) suggests that integrative research, which allows paradigms to be used conjointly, is necessary because integration can lead toincreasingly comprehensive theories, a consequence favored by science. Those in the pluralistic camp disagree. They claim that such approachesare, in, effect, a succumbing to the dominant paradigms, such as the functionalist-positivistic. Jackson and Carter (1991) defend paradigmincommensurability because doing so expands dramatically the scope of organization studies, the interests represented, and those empowered tospeak (p. m). I support the expansion. My argument is that when guided by pluralism, it does not go far enough or deep enough because the necessityfor implementable knowledge is reduced. Therefore, the plea for incommensurability, which is intended to avoid domination by any one approach orcombination of approaches, may have the unintended consequence of becoming a plea for merely a different kind of scientistic authoritarianism.

15.2.4 The Importance of Internal and External ValidityAnother fundamental rule is that the empirical research should pass the tests of internal and external validities. The former stipulates that the researchis designed to minimize unrecognized contaminations and distortions. The latter stipulates that the empirical results can be shown to be relevant incontexts other than those in which the research was conducted.

Elsewhere, I have tried to show the unintended consequences for producing valid actionable knowledge when we focus our criteria upon internal andexternal validity. For example, in order to produce internal validity the experimenter keeps the true objectives of the research secret; controls thechoices the subjects can make; makes undiscussable the design and purpose of the experiment, makes the undiscussability undiscussable, as wellas the modes of analysis to be used. All these conditions are consistent with the top-down relationship between superiors and subordinates intraditional hierarchical organizations (Argyris 1980, 1990, 1993).

If this is true, then the results of experiments should be crafted so that they read that the propositions produced are valid when the conditions used toproduce them are consistent with traditional hierarchical arrangements. To my knowledge, such statements are rarely included. They are, I suggest,crucial if one has implementability in mind.

For example, Barker, Dembo, Lewin (1941), in a carefully designed experiment reported a curvilinear relationship between creativity and frustration.Let us assume that a leader of a group wishes to produce mild degrees of frustration in order to mobilize a desired degree of creativity. How will she doit? The research studies suggest that she should block the group's action. How will she block action? How will she know when the frustration is mild?How will she know when frustration becomes counterproductive? The researchers can specify how she could find out, because the researchers facedthe same challenge. They developed several different types of measuring instruments that they used behind a one-way vision screen. But how will thegroup leader introduce the use of these instruments?

Should she tell the group members about her intentions and about the measuring instruments? How would she interrupt the flow without confoundingher problem? Could not the very mention of her plans increase the frustration?

If there were a way to cover up her strategy, then she would have to cover up the cover-up. What are the ethical issues of researchers producingknowledge that requires creating such undiscussable tactics? What would happen to the sense of trust between the leader and her group if they ever

1



found out? Indeed, what would happen to the credibility of the researchers in the eyes of society?

15.3 Critical Theory and ImplementationSimilar problems arise with scholars who place action and change high on their agenda yet their propositions are not crafted to be implementable. Thefirst example is critical theorists, especially Habermas, and his claims about authentic communication. Scholars associated with critical theory raisequestions about normal science views on causality and the methods used to generate knowledge. Moreover, many of them espouse a view thatinquiry and research should go beyond focusing on the status quo and to generate liberating alternatives.

Geuss (1981:76) proposes that a critical theory is composed of three main constituent parts:

A part which shows that a transition from the present state of society to some proposed final state is possible A part which shows that the transition is practically necessary, i.e. that

1. the present social arrangements cause pain, suffering, and frustration agents only accept the present arrangements becausethey hold a particular world picture one they acquired only because they were in conditions of coercion;2. the proposed final state will be one which will lack the illusions and unnecessary coercion and frustration which will be easier for theagents to realize their true interests;

A part which asserts that the transition from the present state to the proposed final state can come about only if the agents adopt the critical theoryas their self-conscious and act on it.

In order to satisfy the several claims Geuss describes, a critical theory must make many empirical assertions. On the one hand, for example, it mustidentify the causal links between present social arrangements and their negative consequences, and show that alternative arrangements would notcreate these same consequences. On the other hand, it is not clear that the claim that agents would reject their current world view if they were in aposition to know better is empirical. Assuming this claim is not simply a tautology (if they had different wants, they would want different things), what isits status? And related to this, what are we to make of the reference to agents true interests? Assuming agents do not now know better and areunaware of their true interests, how are these to be determined? Geuss identifies the point at issue when he observes that a critical theory doesn'tmerely give information about how it would be rational for agents to act if they had certain interests; it claims to inform them about what interests it isrational for them to have (1981: 58).

Critical theory justifies advocacy of a normative position by adhering to the principle of internal criticism. The ultimate criterion of validity is free assentin a discussion, in which the test of truth is that investigators who begin with different views converge on one opinion in the course of inquiry.

Habermas, for example, argues that human interaction presumes what he calls an ideal speech situation. He argues that all linguistic communicationinvolves four kinds of validity claims: that what is uttered is comprehensible, that the content of what is said is true, that the speaker is being truthful(the utterance is congruent with the speaker's intentions), and that the speech acts being performed are legitimate (Habermas 1979: 2, 28; Bernstein1976: 211). Now, if any of these validity claims are questioned, speakers resort to what Habermas calls a discourse in which the claims areexamined and tested. The criteria for good discourse, that is, for the rationality of the consensus that may be achieved through discourse, are that itapproximates the ideal speech situation. That is, what it means for a statement to be true is that it would be the one on which all agents would agree ifthey were to discuss all of human experience in absolutely free and uncoerced circumstances for an indefinite period of time (Geuss 1981:65).Habermas argues that the ideal speech situation is the grounding for the ideas of rationality, freedom, and justice, as well as the idea of truth.

Habermas has argued that acceptability in the ideal speech situation is a transcendental criterion of truth, by which he means that all human beingseverywhere at all times are committed to recognizing it by virtue of the nature of linguistic communication. This may be a valid claim of how individualsshould act. But, as will be illustrated below, we find that most individuals espouse such commitments but do not act consistently with them. Moreover,they are systematically unaware of the discrepancies and, unaware that they are unaware.

In this connection, I participated in a lengthy seminar at Harvard led by Habermas. The audience was composed primarily of doctoral students, manyof whom, including myself, were admirers of Habermas's contributions. Whenever questions were asked to clarify and inform Habermas actedconsistently with the requirements of the ideal speech situation. However, when the audience raised critical questions neither Habermas nor thestudents acted consistently with the requirements of authentic communication. Nor was this discussable. To my knowledge, Habermas has notconducted inquiries into the problems of implementing the ideal speech situation: how can human beings be educated to diagnose their respectivepositive or negative contributions to producing ideal speech situations?

15.4 Psychoanalysis and ImplementationAnother example of systematic discrepancies may be found in the use of psychoanalysis in studying organizations. This example is especiallyimportant because central to the psychoanalytic clinical perspective is implementation. Recently (Gabriel 1999) and his collaborators published anexcellent analysis of the use of psychoanalytic theory and method in the study of organizations. The authors emphasize that one of the key features ofthe psychoanalytic approach is that it focuses on the features of depth in organizations.

For example, the NASA Challenger disaster is described and psychoanalytic depth explanations are provided. The psychoanalytic explanationconcludes that NASA officials deceived themselves because it was too painful for them to acknowledge that the organization they loved and cared forwas fallible. The blindness of NASA officials, we are told, was caused by repression and narcissism, which led top management to place loyalty



above reality in order to sustain America's narcissistic delusion of itself. These defensive actions were shaped by feelings of pride, anxiety, pain, that,in turn, were connected to their experiences in their early lives. All these causal explanations may be valid but the interpretative processes are notdefined explicitly so that readers, and more important, potential users, can make up their own mind as to its validity and implementability.

How implementable are recommendations based upon the above psychoanalytic explanations? Let us assume that highly skilled psychoanalyticscholars can implement the recommendations. But, how about, for example, the executives whose ways of managing would be greatly affected. It isdoubtful that they could become as skillful as the professionals. It is even more doubtful that they could create an organizational setting where thepsychoanalytic concepts can be used by the participants at all levels of the organization. The psychoanalytic professionals may respond that they donot claim that their perspective is implementable by non-professionals. They recognize that implementable validity for non-professionals may be low.

There are, at least, two problems with this response. First, it makes the practitioners dependent upon the professionals precisely in the area ofmanaging where their sense of competence is highly involved. The resulting sense of unilateral control over the executives, according to thepsychoanalytic theory, should lead to inner conflicts and anxieties, which, in turn, activates defenses on the part of the client. The psychoanalyticprofessionals may be creating the same top-down management action that NASA officials were accused of using and that was diagnosed by thepsychoanalytic professionals as dysfunctional.

The second problem is related to managers developing the requisite skills and using them in the task of managing employees at all levels. All too manyyears ago, I interviewed British executives on organizational issues. They asked me about advice that might emanate from research about intergroupsand interdepartmental conflicts conducted in the United States. I was surprised because one reason I was visiting England was to study the Tavistockapproach to these problems. I asked the executives why they were not looking to Tavistock for help. As I recall, at least six executives, had attendedTavistock workshops. They found them to be interesting. They felt however that British, or any executives, were not likely to develop the requisite skillsnor was it clear how managers or employees could be educated so that legitimate claims of the implementability of the Tavistock concepts could begiven a fair test. The value of psychoanalytic theory, as exemplified by Tavistock, an organization that Gabriel describes as a pioneering one (a claimthat I believe is justified) had decreased in the eyes of the managers.

The reader might ask if it is not a fundamental assumption of psychoanalytic theory that if awareness is enhanced, it will help individuals becomemore effective in designing their actions. The answer is yes. Such an answer, however, raises a deeper question. How do we know when we haveunderstood something? I suggest that our claim that we know something is most rigorously tested if we can produce what we claim we know.

If we claim that through the use of such concepts as neurosis, narcissism, and repression we can explain the Challenger tragedy, a test of this claimwould be for us to produce ways to help NASA learn not to repeat the organizational defensive routines that were documented. In other words,awareness should be in the service of action. Action then becomes the criterion for the validity of our awareness.

This suggests that if other theories exist that provide a causal theory of the Challenger tragedy that is more closely connected to action; to re-educationof the executives, and to organizational change, they should be compared with the psychoanalytic position.

One candidate is the theory of action described in this chapter. An analysis was made of the Challenger disaster using similar reports (Argyris 1990).The analysis suggested that an explanation can be achieved by focusing on the theories of action used by the participants, the organizationaldefensive mechanisms and barriers to effective communication. The analysis also suggested that these problems are not unique to NASA. Indeedmany organizations (private and public) exhibit the same features. Finally, it was shown that this theory has been used to design interventionprograms to correct many of the individual group, and organizational errors described above. Let us now turn to describing the theory of actionapproach.

I should now like to close with a very brief description of how a theory of action approach deals with many of the problems identified above.

15.5 A Theory of Action Approach15.5.1 Designs for Action: Model IHuman beings hold two types of theories of action. There is the one that they espouse, which is usually expressed in the form of stated beliefs andvalues. Then there is the theory that they actually use; this can only be inferred from observing their actions, that is, their actual behavior. To date, mosthuman beings studied have the same theory-in-use. There is diversity in espoused theories, but not in theories-in-use. A model of the theory-in-use (thatwe call Model I) follows. Model I theory-in-use is the design we found throughout the world. It has four governing values. They are: (1) achieve yourintended purpose, (2) maximize winning and minimize losing, (3) suppress negative feelings, and (4) behave according to what you consider rational.

The most prevalent action strategies that arise from Model I are the following: (1) advocate your position, (2) evaluate the thoughts and actions ofothers (and your own thoughts and actions), and (3) attribute causes for whatever you are trying to understand (Argyris 1982, 1990, 1993; Argyrisand Schn 1996).

These actions must be performed in such a way that satisfies the actors governing valuesthat is, they achieve at least their minimum acceptablelevel of being in control, winning, or bringing about any other result. In other words, Model I tells individuals to craft their positions, evaluations, andattributions in ways that inhibit inquiries into and tests of them with the use of independent logic. The consequences of these Model I strategies are likelyto be defensiveness, misunderstanding, and self-fulfilling and self-sealing processes (Argyris 1982).

Model I theory-in-use requires defensive reasoning. Individuals keep their premisses and inferences tacit, lest they lose control. They create tests oftheir claims that are self-serving and self-sealing. The likelihood of misunderstanding and mistrust increases. The use of defensive reasoning prohibits

2



questioning the defensive reasoning. We now have self-fueling processes that maintain the status quo, inhibit genuine learning, and reinforce thedeception. (Figure 15.1)

Human beings learn their theories-in-use early in life, to use them constantly, and therefore the actions that they produce are highly skilled. Littleconscious attention is paid to producing skilled actions. Indeed, conscious attention could inhibit producing them effectively. This creates unawarenessof what we are doing when we act skillfully. The unawareness due to skill and the unawareness caused by our unilaterally controlling theories-in-useproduce a deeper unawareness; namely, we become unaware of the programs in our heads that keep us unaware. The results are skilledunawareness and skilled incompetence (Argyris 1980). For example, when individuals have to say something negative to others (e.g. Yourperformance is poor) they often ease-in, in order not to upset the other. Two of the most frequent easing-in actions that we observe are (a) non-directivequestioning and (b) face-saving approaches. In order for these to work, the individuals must cover up that they are acting as they are, in order not toupset the other. In order for a cover-up to work, the cover-up itself must be covered up.

Under these conditions, we find that the recipients are wary of what is happening. They sense that there may be a cover-up. Because they hold thesame theory-in-use, they too cover up their private doubts. The result is counterproductive consequences for genuine problem-solving. All of this occurswith the use of skillful behavior; hence, the term skilled incompetence.

When organizational worlds become dominated by these consequences, human beings become cynical about changing the self-fuelingcounterproductive process.

Click to view largerFig. 15.1 Model I theory-in-use

Not surprisingly, they learn to distance themselves from taking responsibility, losing, and suppressing negative feelings, especially those associatedwith embarrassment or threat. Individuals use behavioral strategies consistent with these governing values. For example, they advocate their views,making evaluations and attributions in such a way as to ensure their being in control, winning, and suppressing negative feelings.

In short, individuals learn theories-in-use that are consistent with producing unilateral control. It is true that organizations are hierarchical and based onunilateral control. It is equally true that individuals are even more so. Place individuals in organizations whose structures are designed to be moreegalitarian, and individuals will eventually make them more unilateral and authoritarian. The most massive examples of such situations of which I amaware are the alternative schools and communes of the 1970s. Most have failed and slowly have faded away (Argyris 1974).

15.5.2 Organizational Defensive RoutinesOrganizational defensive routines are any action, policy, or practice that prevents organizational participants from experiencing embarrassment orthreat and, at the same time, prevents them from discovering the causes of the embarrassment or threat. Organizational defensive routines, like ModelI theories-in-use, inhibit genuine learning and overprotect the individuals and the organization (Argyris 1990).

There is a fundamental logic underlying all organizational defensive routines. It can be illustrated by one of the most frequently observed defenses,namely, sending mixed messages, such as, Mary, you run the department, but check with Bill, or lohn, be innovative but be careful. The logic is: (1)send a message that is inconsistent, (2) act as if it is not inconsistent, (3) make Steps 1 and 2 undiscussable, and (4) make the undiscussabilityundiscussable.

Organizational defensive routines are caused by a circular, self-reinforcing process in which individuals Model I theories-in-use produce individualstrategies of bypass and cover-up, which result in organizational bypass and cover-up, which reinforce the individuals theories-in-use. The explanationof organizational defensive routines is therefore individual and organizational. This means that it should not be possible to change organizationalroutines without changing individual routines, and vice versa. Any attempts at doing so should lead to failure or, at best, temporary success.

15.5.3 Model II: Theories-in-useTo help individuals recognize their skillful Model I blindness, an intervener introduces Model II theories-in-use. Model II theories are, at the outset,espoused theories. The challenge is to help individuals transform their espoused theories into theories-in-use by learning a new set of skills and a newset of governing values. Because many individuals espouse Model II values and skills, these traits are not totally new to them. However, the empiricalfact to date is that very few individuals can routinely act on their espoused values, and they are often unaware of this limitation.

The governing values of Model II are valid information, informed choice, and vigilant monitoring of the implementation of the choice in order to detectand correct error. (Figure 15.2) As in the case of Model I, the three most prominent behaviors are advocate, evaluate, and attribute. However, unlikeModel I behaviors, Model II behaviors are crafted into action strategies that openly illustrate how the actors reached their evaluations or attributions,and how they crafted them to encourage inquiry and testing by others. Productive reasoning is required to produce such consequences. Productivereasoning means that the premisses are explicit, the inferences from the premisses are also made explicit, and finally conclusions are crafted in waysthat can be tested by logic that is independent of the actor. Unlike the defensive reasoning, the logic used is not self-referenced. As a consequence,defensive routines that are anti-learning are minimized, and genuine learning is facilitated. Embarrassment and threat are not bypassed and covered



up; they are engaged (Argyris 1982; Argyris and Schn 1974).

To the extent that individuals use Model II theory instead of merely espousing it, they will begin to interrupt organizational defensive routines and beginto create organizational learning processes and systems that encourage double-loop learning in ways that persist. These are called Model II learningsystems (Argyris and Schn 1996).

Click to view largerFig. 15.2 Model II theory-in-use

15.5.4 Design of the Research-Intervention ActivitiesThere are design rules that follow from the theoretical framework described above that can be used to design the research and the interventionactivities:

Discover the degree to which the clients theories-in-use are consistent with Model I. Discover the degree to which the clients use defensive reasoning whenever they deal with embarrassing or threatening issues. Discover the designs (rules) the clients have in their heads that keep them unaware of the discrepancies among their espoused values, theiractions, and their theories-in-use. Discover the degree to which the clients discourage valid reflection on their actions while they are acting. To put this another way: discover how theclients create designs for action that they do not follow but that they believe they do follow, while they are also being systematically unaware of thisdiscrepancy and are behaving in ways that prevent them from discovering the discrepancy and the causes of their unawareness. Discover the defensive routines that exist in the organization and limit learning. Develop maps of these organizational defensive routines,specifying the actions that lead to limited-learning consequences and cause them to persist even though the directors wish to be free of them.

In order to reach these goals, re-education and change programs begin by producing relatively directly observable data about these clients reasoningand actions. The clients must accept responsibility for creating these data, and these data must be in a form from which the clients theories-in-use canbe inferred (e.g. recorded conversations):

Encourage the clients to examine inconsistencies and gaps in the reasoning that underlines their actions. Surface and make explicit the rules that must be in their heads if they maintain there is a connection between their designs for action and theactions themselves. View any bewilderment or frustration that results as further directly observable data that can be used to test the validity of what is being learned. Produce opportunities to practice Model II ways of crafting actions that will reduce counterproductive consequences.

In principle, the kind of research of which I write can begin with identifying either the theories-in-use or the organizational defensive routines. It does notmatter which, because one will necessarily lead you to the other. I usually make the choice on the basis of which of the two is most likely to generate theparticipants internal commitment to the research and to the eventual intervention.

15.5.5 The Left- and Right-hand Column Case MethodThe case method described next is one of several instruments used in our action science research (Argyris, Putnam, and Smith 1985; Argyris andSchn 1996). The key features of all the research methods and this case method in particular are:

1. It produces relatively directly observable data such as conversation. Such data are the actual productions of action, and therefore canbecome the basis for inferring theories-in-use.2. It produces data in ways that make clear the actors responsibility for the meanings produced. When used properly, the respondents cannotmake the research instrument causally responsible for the data that they produced (e.g. I didn't really mean that; or I didn't understand themeaning of that term).3. It produces data about the respondents causal theories, especially those that are tacit because they are taken for granted.4. It provides opportunities for the respondents to change their responses without hindering the validity of the inferences being made. Indeed, theactions around changing their minds should also provide data about their causal reasoning processes. It provides opportunities to change theiractions as well as actions of groups, intergroups, and organizations over which they have some influence. It provides such knowledge in waysthat are economical and do not harm the respondents or the context in which they are working.

The directions to write a case are given to each individual. The directions request:

1. In one paragraph, describe a key organizational problem as you see it.2. Assume you could talk to whomever you wish in order to begin to solve the problem. Describe, in a paragraph or so, the strategy that you



would use in this meeting.3. Next, split your page into two columns. On the right-hand side write how you would begin the meetingwhat you would actually say. Then writewhat you believe the other(s) would say. Then write your response to their response. Continue writing this scenario for two or so double-spacedtypewritten pages.4. In the left-hand column, write any idea or feeling that you would have that you would not communicate for whatever reason.

In short, the case includes a statement of the problem, the intended strategy to begin to solve the problem, the actual conversation that did or wouldoccur as envisioned by the writer, the information that the writer would not communicate for whatever reason.

15.5.6 Reflecting on the CasesIn analyzing their left-hand columns, to cite one example, the executives found that each side blamed the other side for the difficulties, and they usedthe same reasons. For example, each side said about the other side: You do not really understand the issues If you insist on your position, you willharm the morale that I have built. Don't hand me that line. You know what I am talking about. Why don't you take off your blinders and wear acompany hat?

These results illustrate once more the features of skilled incompetence. It requires skill to craft the cases with the intention not to upset others whiletrying to change their minds. Yet, as we have seen, the skilled behavior used in the cases had the opposite effect. The others in the case becameupset and dug in their heels about changing their minds.

15.5.7 Redesigning their ActionsThe next step is to begin to redesign their actions. The executives turned to their cases. Each executive selected an episode that he wished to redesignso that it would not have the negative consequences. As an aid in their redesign, the executives were given some handouts that described Model II setof behaviors. The first thing they realized was that they would have to slow things down. They could not produce a new conversation in the millisecondsthat they were accustomed. This troubled them a bit, because they were impatient to learn. They kept reminding themselves that learning new skillsdoes require that they slow down.

One technique they used was that each individual crafted, by himself, a new conversation to help the writer of the episode. After taking five or sominutes, the individuals shared their designs with the writer. In the process of discussing these, the writer learned much about how to redesign hiswords. However, the designers also learned much as they discovered the gaps in their suggestions and the ways in which they made them.

Practice is important. Most people require as much practice to learn Model II as is required to play a not-so-decent game of tennis. However, thepractice does not need to occur all at once; it can occur in actual business meetings where people set aside some time to make it possible to reflect ontheir actions and correct them. An outside facilitator could help them examine and redesign their actions, just as a tennis coach might do. But, as in thecase of a good tennis coach, the facilitator should be replaced eventually by the group. The facilitator might be brought in for periodic boosters or tohelp when the problem is of a degree of difficulty and intensity not experienced before.

There are several consequences of this type of change program. First, the executives begin to experience each other as more supportive andconstructive. People still work very hard during meetings, but their conversation begins to become additive; it flows to conclusions that they all can ownand implement. Crises begin to be reduced. Soon, the behavioral change leads to new values, and then to new structures and policies to mirror thenew values (Argyris 1984, 1993; Argyris and Schn 1996). This, in turn, leads to more effective problem-solving and decision-making. In the case ofthis group, they were able to define the questions related to strategy, to conduct their own inquiries, to have staff people conduct some relevantresearch, and to have three individuals organize it into a presentation that was ultimately approved and owned by the top group. The top group alsobuilt in a process of involving their immediate reports so that they too could develop a sense of ownership, thereby increasing the probability that allinvolved will work at making it successful.

15.5.8 Basic Criteria for Success in Diagnosing and Changing at any Level of the OrganizationThere are four criteria that are central to design diagnostic instruments and interventions in organizations. They are:

1. The criterion for ultimate success should not be change in behavior or attitudes. The criterion should be changes in defensive reasoning and thetheories-in-use that produce skilled unawareness and skilled incompetence and the resulting organizational defensive routines.2. The changes just described should unambiguously lead to reductions in the self-fulfilling counterproductive activities, at all levels of theorganization.3. It is not possible to achieve Criteria 1 and 2 without focusing on the actual behavior of the participants. The trouble with the old criteria is thatthey began and ended with behaviors. The new criteria begin with behavior in order to get a window into the mental maps and type of reasoningthat the individuals use and the organizational culture that they create.4. The success of programs is not assessed by measuring insight gained or learning reported by the participants. Individuals often report highscores on insight and learning, yet have not changed their defensive reasoning, their theories-in-use, their skilled unawareness and incompetence,and the organizational defensive routines.

15.5.9 Summary1. Actionable knowledge requires propositions that make explicit the causal processes required to produce action. Causality is key in implementation.Shoham (1990) writes, If causal reasoning is common in scientific thinking, it is downright dominant in every day sense making (p. 214).



2. One of the most powerful inhibitors of effective action is inner contradictions. Inner contradictions exist when the propositions to act are implementedcorrectly, the result is the effective consequence predicted and necessarily counterproductive consequences that were not predicted.

3. One cause of inner contradiction is the methodologies used by most normal social scientists to discover problems and to invent solutions. Thedesign of these methodologies is dominated by the requirements of internal and external validity.

4. The features above cause the degree of seamlessness and the validity of the implementation to be reduced. Moreover, the actors tend not to beaware of these problems while they are producing them. If they become aware, the tendency is to blame factors for which they claim they are notresponsible.

5. The focus on describing reality in ways that satisfies the requirements of internal and external validity makes it less likely that attention is paid to theimplementable validity of the propositions. This, in turn, leads to propositions that are abstract (in order to be generalizable) and disconnected fromimplementable action. This is a major problem because much of the advice about effective leadership, organizational learning and transformationalchange promises Model II consequences but they are formulated in Model I concepts. This results in major inconsistencies, which, in turn, results in theadvice losing its credibility, and is often relegated the status of fads (Argyris 2000).

15.6 Implications for Future Research1. It is not news for social scientists that they must master the current ideas about producing valid knowledge. For example, the concepts of internaland external validity are well known and respected. What may be news is that following the rules for producing internal and external validity will tend tocreate gaps, inconsistencies, and inner contradictions, especially when the intent is to produce actionable knowledge that questions the status quoand to create rare emancipatory events. Moreover, as we have seen the same rules for conducting rigorous empirical research will tend to produceunawareness of the gaps, inconsistencies, and inner contradictions.

The reason is that actionable knowledge about effective action must be formulated in the form of causal propositions that are storable, retrievable, andimplementable by the executive function of the mind/brain. In order for this to be accomplished the propositions must be normative (because thecriteria for effectiveness are based on subjective or personal choices). They must also be formulated in the form of prescriptions for action that theexecutive function can implement. These are the very features that social scientists are told to downplay.

Both positivists and humanists exhibit the same problem. For example, positivists may specify an empirical relationship between two variables. Thisis accomplished by developing systematic observable patterns and then describing these quantitatively. But such propositions do not specify thecausal designs required by the executive function to produce this relationship. The humanists who focus on crafting stories, seeking authenticity incommunication, exhibiting empathy with the subjects and emphasizing wholeness have not to my knowledge, specified the causal designs that theexecutive function requires if it is to implement these features.

2. Policies, plans, strategies are not actionable knowledge. The gaps and inconsistencies identified above also exist at a more macro level. I refer tothose who craft policies, plans, and strategies. Such plans are inventions for effective action but not designs for effective implementation of theinvention. In order to explain the reasoning behind this claim, I should like to present a model (largely based on the work of Dewey and Lewin).

In order to produce effective action, human beings implement four phases of a cycle for designing and executing action. The phases are to discover aproblem, to invent a solution, to produce the invented solution, and to evaluate the production (Figure 15.3).

Click to view largerFig. 15.3 The Action Cycle



Click to view largerFig. 15.4 The action and sub-routine cycle

Figure 15.1 does not adequately describe what is going on in order to produce action. The reason is that action is required in order to implement eachof the phases. For example, the phase discovery, if it is to be produced, requires actions that include discover how to discover, invent ways todiscover, produce the inventions, and evaluate the effectiveness of the production. This is true for invention, production, and evaluation. If they are to beimplemented each requires action and therefore each requires sub-cycles to produce their respective phases. We have therefore, a model similar to ahologram where each part (sub-cycle) of the whole contains the same features of the whole (Figure 15.4).

Actionable knowledge is produced when all four phases and their respective sub-cycles are implemented. It is important not to view the knowledgerequired to produce discovery and invention as actionable. These actions are in the service of producing discovery and inventions and not inproducing the intended consequences derived from the inventions. The moment that attempts are made to implement the plans or policies, thoseactions are about actionable knowledge.

For example, two recent and well-known examples, in the United States, are the Clintons health programs and the policy recommendations forsecuring nuclear secrets defined by the Secretary of Energy, Mr Richardson. Both plans failed the criterion of being implemented. The creators of thehealth policy assumed that if their plans were first rate they would be implemented. The Secretary acknowledged that the security plans were notimplemented because of what he described as the culture of the laboratories.

3. Researchers will have to become more concerned about their credibility with the society at large that funds and wishes to use their research. Forexample, in the United States, a congressman who was a strong supporter for giving social scientists freedom to choose their research, reluctantlyconcluded, after twenty years, that he was systematically lied to by social scientists who promised that user-friendly actionable knowledge was aprimary concern (Johnson 1993). In the early days, the Ford Foundation had a similar policy of encouraging the freedom of social scientists to choosetheir research. After several decades, the officials decided if the Foundation were to help mankind then they would have to define the research to bedone. Indeed many now support various experiments designed and produced by non-social scientists.

4. Social psychologists, sociologists, and anthropologists have produced an enormous literature about the impact that situations, roles, and culturehave upon the way human beings think and act; upon their sense of competence, self-efficacy, self-confidence, and upon their fundamental values.May I suggest that we will take seriously these findings to examine our actions and our respective scientific communities. I suggest that we create forourselves, what Thorstein Veblen called trained incapacity (1919).

Imagine, for example, what would happen if social scientists dealt in the way they deal with those seeking to benefit from their research. They mightsay to the child facing conflict or bewilderment, in effect, that they focus on describing the universe. For example, my focus is describing phenomena,and I cannot offer any suggestions to you. Finally imagine a child pleading for prescriptions and being told by the parents that the existing descriptivedata are inadequate to formulate prescriptions. This is precisely what happens when adults called clients, seek help from those scholarly-consultants,who adhere to the rules described above. In a study that I have just begun, I have interviewed and observed world-class consultants who are alsoworld-class scientists. They all hold, with various degrees of explicitness, a framework to understand and explain as well as rigorous empiricalmethods to conduct inquiry. I ask them if they have been in a situation where they felt their analysis was sound and the clients expressed doubts. Allresponded affirmatively. Next, I ask what do they do. In effect they repeat the presentation to assure the client of its validity. They often ask the clients topresent the reasoning behind their doubts. When they do, the scholar-consultants identify gaps and inconsistencies in the clients reasoning processesby using the framework that got them into trouble in the first place. They appear to hold the rule: if the client doubts the validity of the presentation,repeat the presentation. If the client continues to have doubts, encourage them to express the doubts and, by using the original framework, help them tosee where they are wrong. If repetition of this strategy does not work, find polite ways to exit the relationship. By the way, the clients also produce theirversion of defense. The two versions clash and the result is little learning and self-sealing processes.

Holding a position that makes the production of normative and prescriptive knowledge as central makes social scientists vulnerable. It is one thing tobe challenged about actionability and be able to respond that one is focused on being descriptive. It is quite another to be challenged when the socialscientists prescriptions do not work as predicted. Perhaps the defense of this trained incapacity is an additional cause for the down play ofactionable knowledge.

ReferencesAllison, G. (1971). Essence of Decision: Explaining the Cuban Missile Crisis. Boston: Little, Brown.

Argyris, C. (1980). Inner Contradictions of Rigorous Research. San Diego: Academic Press.



(1982). Reasoning, Learning and Action: Individual and Organizational. San Francisco: Jossey-Bass.

(1990). Overcoming Organizational Defenses: Facilitating Organizational Learning. Needham, Mass.: Allyn & Bacon.

(1993). Knowledge for Action. San Francisco: Jossey-Bass.

(1995). Knowledge When Used in Practice to Test Theory: The Case of Applied Communication Research, in Kenneth N. Cissna (ed.), AppliedCommunications in the 21st Century. Mahwah, NJ: Lawrence Erlbaum.

(1997). Field Theory as a Basis for Scholarly Consulting. Journal of Social Issues, 53/4: 80924.

(2000). Flawed Advice and the Management Trap. New York: Academic Press.

and Kaplan, R. S. (1994). Implementing New Knowledge: The Case of Activity-Based Costing. Accounting Horizons, 8/3: 83105.

and Schn, D. (1974). Theory in Practice. San Francisco: Jossey-Bass.

(1996). Organizational Learning II. Reading, Mass.: Addison Wesley.

Putnam, R., and Smith, D. (1985). Action Science: Concepts, Methods, and Skills for Research and Intervention. San Francisco: Jossey-Bass.

Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall.

Barker, R. G., Dembo, T, and Lewin, K. (1941). Frustration and Regression. University of Iowa Studies in Child Welfare 1. Ames: University of IowaPress.

Bernstein, R. J. (1976). The Restructuring of Social and Political Theory. Philadelphia: University of Pennsylvania Press.

Bolman, L. G., and Deal, T. E. (1991). Reframing Organizations: Artistry, Choice and Leadership. San Francisco: Jossey-Bass.

Burrell, G., and Morgan, G. (1979). Sociological Paradigms and Organizational Analyses. London: Heinemann.

Coleman, J. S. (1972). Policy Research in the Social Sciences. Morristown, NJ: General Learning Press.

Cronbach, L. J., and Suppes, P. (eds.) (1969). Research for Tomorrow's Schools. London: Macmillan.

Cyert, Richard M., and March, James G. (1963). A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall.

Donaldson, L. (1985). In Defense of Organizational Theory. Cambridge: Cambridge University Press.

Dyer, G. W., Jr, and Wilkins, A. L. (1991). Better Stories, not Better Constructs, to Generate Better Theory: A Rejoinder to Eisenhardt. Academy ofManagement Review, 16/3: 61319.

Eisenhardt, K. M. (1989). Building Theories from Case Study Research. Academy of Management Review, 14/4: 53250.

Ellis, P. (1982). The Phenomenology of Defensible Space, in P. Stringer (ed.), Confronting Social Issues. San Diego: Academic Press.

Engstrm, Yrj (2000). From Individual Action to Collective Activity and Back: Developmental Work Research as an Interventionist Methodology, inPaul Luff, John Hind-marsh, and Christian Heath (eds.), Workplace Studies. Cambridge: Cambridge University Press.

Gabriel, Yiannis, et al. (1999). Organizations in Depth: The Psychoanalysis of Organizations. London: Sage Publications.

Geuss, R. (1981). The Idea of a Critical Theory. Cambridge: Cambridge University Press.

Giddens, A. (1976). New Rules of Sociological Method. London: Hutchinson.

Gill, J., and Johnson, P. (1991). Research Methods for Managers. London: Chapman.

Habermas, J. (1979). Communication and the Evolution of Society. Boston: Beacon Press.

Hassard, J. (1991). Multiple Paradigms and Organizational Analysis: A Case Study. Organization Studies, 12/2: 27599.

Jackson, N., and Carter, P. (1991). In Defense of Paradigm Incommensurability. Organization Studies, 12/1:10927.

Johnson, D. (1993). Psychology in Washington. Measurement to Improve Scientific Productivity: A Reflection on the Brown Report. PsychologicalScience, 4/2: 679.

Lewin, K. (1935). A Dynamic Theory of Personality. New York: McGraw-Hill.

(1948). Resolving Social Conflicts, ed. G. W. Lewin New York: HarperCollins.

(1951). Field Theory in Social Science, ed. D. Cartwright. New York: HarperCollins.

Mcgrath, J. E., and Brinberg, D. (1984). Alternative Paths for Research. Applied Social Psychology Annual, 5:10920.

Martin, J., and Kukla, R. A. (1982). Judgment Calls in Research. Newbury Park, Calif.: Sage.



March, James G. (1981). Decision-Making Perspective, in Andrew H. Van De Ven and William F. Joyce (eds.), Perspectives on Organizational Designand Behavior. New York: John Wiley & Sons.

(1988). Decisions and Organizations. Oxford: Basil Blackwell.

Morgan, G. (1983). Beyond Method. Newbury Park, Calif.: Sage.

and Smircich, L. (1980). The Case for Qualitative Research. Academy of Management Review, 5/4: 491500.

Peters, M., and Robinson, V. (1984). The Origins and Status of Action Research. Journal of Applied Behavioral Science, 20/2:11324.

Rosen, M. (1991). Coming to Terms with the Field: Understanding and Doing Organizational Ethnography. Journal of Management Studies, 28/1:124.

Scott, W. R. (1981). Organizations: Rational, Natural and Open Systems. Englewood Cliffs, NJ: Prentice-Hall.

Seashore, S. E. (1985). Institutional and Organizational Issues in Doing Useful Research, in E. E. Lawler III, A. M. Mohrman, Jr., S. A. Mohrman, G. E.Ledford, Jr., T. G. Cummings, and Associates (eds.), Doing Research that is Useful for Theory and Practice. San Francisco: Jossey-Bass.

Shoham, Y. (1990). Nonmonotonic Reasoning and Causation. Cognitive Science, 14: 213302.

Simon, H. A. (ed.) (1976), Administrative Behavior: A Study of Decision-Making Processes in Administrative Organization. New York: Free Press.

(1990). Invariants of Human Behavior. Annual Review of Psychology, 41:120.

Stringer, P. (ed.) (1982). Confronting Social Issues, ii. San Diego: Academic Press.

Van Maanen, J. (1982). Introduction, in J. Van Maanen, J. M. Dabbs, Jr, and R. R. Faulkner (eds.), Varieties of Qualitative Research. Newbury Park,Calif.: Sage.

Veblen, Thorstein (1919). The Theory of the Leisure Class. New York: B. W. Hurbach.

Wheelan, S. A., Pepitone, E. A., and Abt, V. (1990). Advances in Field Theory. Newbury Park, Calif.: Sage.

White, R. W. (1959). Motivation Reconsidered: The Concept of Competence. Psychological Review, 66: 297333.

Whyte, W. F. (1991). Social Theory for Action: How Individuals and Organizations Learn to Change. Newbury Park, Calif: Sage.

Notes:(1) Because of space limitations, I refer to my book The Inner Contradictions of Rigorous Research for many more examples.

(2) Because of space limitations I am able to select an example that deals with individual and group phenomena. Illustration of organizationalphenomena may be found in Argyris 1982, 1990, 1993, Argyris and Schn 1996.

Chris ArgyrisChris Argyris is the James Bryant Conant Professor Emeritus, Harvard University. His books include Overcoming OrganizationalDefenses (1990), Knowledge for Action (1993), Flawed Advice (2000), Reasons and Rationalizations (2004), Theory in Practice (1994),and Organizational Learning II (1996), both with Donald Schn, and Action Science (1985) with Diana Smith and Robert Putnam.



actionable knowledge

Documents